After approval by the Ethics Committee of the University Hospital of Zurich and written informed consent from all participating individuals, blood samples for DNA extraction were obtained from Caucasian patients with ICP or CIC. The total population of analyzed individuals consisted of two different groups: 25 patients (21 ICP_new patients and four CIC patients) were prospectively recruited for this study, and a second group of 20 patients (ICP_old) had already been described in a previous study by Pauli-Magnus and coworkers[11].

Two hundred and five Caucasian volunteers and patients without cholestasis, as well as Caucasian women with uneventful pregnancies (n = 40), served as a control population for BSEP (ABCB11) and MDR3 (ABCB4) genetic variants. These controls have already been described in previous studies[112526]. Specifically, pregnant controls were all healthy, as defined by normal serum levels of transaminases, bilirubin, γ-GT, alkaline phosphatase (AP) and bile acids. Caucasian controls from the study of Pauli-Magnus and coworkers[26] (n = 95) were healthy volunteers recruited for participation in phaseIstudies, with uneventful medical history and normal blood biochemistry. Neither of these two control groups took any regular medication. In the case of the Caucasian control population of Meier and coworkers[25] (n = 110), most patients suffered from extrahepatic malignancies, and cholestatic disease was excluded in all patients. Furthermore, none of these patients used medication known to be associated with the development of cholestasis.

For lack of DNA availability, only 110 out of 205 Caucasian controls could be used for MRP2 sequencing. For the same reason, a new group of Caucasian women with uneventful pregnancies (n = 42) had to be collected for the MRP2 variants. Demographic data and pregnancy course of these women did not differ from the previous control group.

Diagnosis of ICP was based upon: (1) a clinical history of pruritus, which occurred in the third trimester of pregnancy; (2) the presence of laboratory abnormalities suggestive of ICP: fasting serum bile acid ≥ 1.5 ULN (upper limit of normal) and/or serum AP levels ≥ 1.5 ULN and/ or alanine aminotransferase (ALT) levels ≥ 1.5 ULN; and (3) spontaneous resolution of clinical symptoms and laboratory findings after delivery. Diagnosis of CIC was based upon laboratory abnormalities as defined for ICP and the exclusion of preexisting liver disease defined by: (1) a negative serology for hepatitis A, B and C; (2) the exclusion of other preexisting medical conditions that could explain liver injury, such as congestive heart failure, systemic infection, or malignancy; (3) normal liver ultrasound; and (4) a clear causal relationship to drug intake. Each case of ICP and CIC was evaluated by at least one obstetrician and one hepatologist, as well as by a clinical pharmacologist.

Full length ABCB4 and ABCB11 sequencing data were already available from the control groups, as well as from ICP_old patients. To allow detection of additional ABCB4 and ABCB11 mutations in the new group, complete sequencing of these two genes was also performed in the 25 newly recruited patients. Genotyping of ABCC2 included all CIC patients, as well as 17 out of 21 patients from the ICP_new group and 16 of 21 patients in the ICP_old group, which yielded a total number of 33 patients for ABCC2 genotyping. In nine patients (four ICP_new and five ICP_old), no ABCC2 genotyping could be performed for lack of DNA availability.

Isolation of DNA and DNA sequencing was done at Epidauros Biotechnology AG, Bernried, Germany. Genomic and cDNA sequences were derived from known sequences (ABCB4: AC005068.2 for non-coding exons -3 to 1 and coding exons 2 and 3; AC006154.1 for exons 4 to 12; AC0005045.2 for exons 13 to 28; and NM_000443.2 for cDNA; ABCB11: GenBank accession number AC008177.3 for promoter and exons 1 to 21; AC069165.2 for exons 22 to 28 and NM_003742.2 for cDNA).

ABCB4 and ABCB11: Sequencing of ABCB4 covered a proximately 8000 bp, including (1) 2000 bp of the upstream promoter region and non-coding exon -3 to 1 and, (2) coding exons 2-28 and 100-350 bp of the intronic sequence around each exon. For ABCB11, sequencing covered 10 000 bp including (1) non-coding exon 1 and 2400 bp of the upstream promoter region and, (2) coding exons 2-28 and 100-350 bp of the intronic sequence around each exon. Primers for genomic DNA were designed to span all exons and at least 100 bp of the flanking intronic sequence at the 5’ and 3’ end of each exon. The DNA sequence of purified PCR fragments was analyzed on an ABI3700 capillary sequencer (ABI, Weiterstadt, Germany) and assembled using the phredPhrap, Consed and PolyPhred software (University of Washington). Details regarding the primers, optimized PCR conditions and subsequent purification and sequencing of the fragments are available at info@epidauros.com.

ABCC2: Three non-synonymous polymorphisms with a potential impact on MRP2 function and expression were chosen for genotyping[25]: 1249G>A variant (V417I, rs2273697), 3563T>A (V1188E, rs17222723) and 4544G>A (C1515Y, rs8187710). Genotyping was performed with the Custom TaqMan SNP Genotyping Assays procedure (Applied Biosystems, Foster City, CA, USA) which contained a sense- and an antisense primer and two probes, labeled with fluorescent reporter dyes, either VIC or 6-Fam at the 5’ end and a non-fluorescent quencher at the 3’ end to distinguish between alleles 1 and 2, respectively. Primer and probe sequences for individual SNPs are given in Table 1. Probe solution (0.625 &mgr;L) and 12.5 &mgr;L of 2 × Universal PCR Master Mix (Applied Biosystems) were brought to 25 &mgr;L with 20 ng of genomic DNA. PCR reaction (2 min at 50°C, followed by 10 min at 95°C and 40 cycles of 15 s at 92°C and 1 min at 60°C). Allelic discrimination was processed with an ABI PRISM 7700 Sequence Detector.

Genotype distribution, allelic frequencies and odds ratios (ORs) are given with 95% CI. In ABCB11, formal statistical analysis was only performed for the 1331T>C polymorphisms (rs2287622), whereas for ABCC2 analysis, it included two highly linked polymorphisms. No correction according to Bonferroni was, therefore, required. Differences investigated in our study apply to a proportion of diseased versus non-diseased individuals within the whole population, using an unmatched case control design. Response (ICP versus non-ICP) and predictors (T versus C) were both binary variables and were therefore best condensed into a 2 × 2 table. Differences in genotype distribution between patients and controls were calculated with the χ² test, and difference in allelic frequencies between two groups was performed using a 2 × 2 Fisher exact test. P ≤ 0.05 was considered statistically significant.

A total of 25 unrelated patients with estrogen-associated intrahepatic cholestasis were prospectively enrolled in this study, 21 with ICP and four with oral CIC. Demographic data and laboratory findings in ICP patients are given in Table 2. Only two patients showed elevated γ-GT levels > 1.5 ULN, while total bile acid levels were elevated in all patients in whom it was determined (16 out of 21; range, 1.7-17.3 ULN). Three patients had a previous history of ICP; three pregnancies were twin pregnancies, and one patient experienced cholestasis under previous oral contraception.

Characteristics of patients with CIC are given in Table 3. One patient showed elevated γ-GT levels. Total bile acid levels were elevated in all three patients in whom it was determined (three out of four; range, 1.6-22.3 ULN). Oral contraceptive preparations used in the four patients contained comparable amounts of ethinylestradiol (20-35 &mgr;g) while the progesterone-like portion ranged from 50 to 150 &mgr;g. All patients had a liver biopsy done for strictly diagnostic reasons, which showed intrahepatic cholestasis in three patients. One patient had a previous history of ICP.

ABCB4 and ABCB11: Sequence analysis in the 25 newly recruited patients with estrogen-associated cholestasis revealed no disease-associated non-synonymous mutations in ABCB4 or ABCB11. Furthermore, in line with previous findings[11], no ABCB4 polymorphism was found to be overrepresented in the ICP and CIC groups compared to pregnant women without cholestasis and healthy Caucasian individuals. All of the detected genetic variants in ABCB11 and ABCB4 were in Hardy Weinberg equilibrium.

In contrast, the ABCB11 1331T>C → V444A polymorphism was significantly more frequent in ICP and CIC patients compared to the two control groups. Specifically, the CC genotype was encountered in 57.1% of all ICP patients (ICP_new, 47.6% and ICP_old, 67.7%) and 100% of CIC patients compared to 20 and 32.2% in pregnant women without cholestasis and healthy Caucasian controls, respectively (Table 4). In line with these findings, the ORs of C versus T were 3.0 (1.7-6.4) for all ICP patients (ICP_new + ICP_old) versus healthy pregnant control women (ICP_new, 2.1; 1.0-4.7 and ICP_old 4.8; 2.2-15.0) (Table 5 and Figure 1). With the exception of this polymorphism and two intronic variants that were found to be closely linked to the 1331T>C polymorphism in previous studies [intron 13: (+70) C>T and intron 14 (+32) T>C][26], the allele frequency of the remaining common variants in the patients with ICP and CIC was comparable to that observed in healthy pregnant and Caucasian controls. Due to the small sample size, no significance levels could be calculated for the CIC group. However, all patients in this group were homozygous for the C at position 1331, which is highly suggestive of an overrepresentation of this allele compared to the control groups.

Open in New Tab   Full Size Figure   Download Figure

Figure 1 Allelic frequency of the T allele (white panel) and C allele (black panel) of the ABCB11 1331T>C (1331T>C) polymorphism. 21 ICP_new patients (42 alleles); 21 ICP_old patients (42 alleles); 42 ICP_total patients (84 alleles); 20 ICP_control patients (40 alleles); 205 Caucasian_control patients (410 alleles). Allelic frequencies and ORs are given with 95% CI. Groups were compared with Fisher's exact test.

ABCC2: The 1249G>A polymorphism was not found in our patients. In contrast, 3563T>A and 4544G>A were strongly linked and distributed similarly in all groups (Table 4). No significant difference in the frequency of these two polymorphisms was observed between affected ICP and CIC patients and healthy controls. Heterozygous carriers of the 3563A and the 4544A alleles were found in 15.2% of ICP patients (ICP_new, 23.5% and ICP_old, 3.1%) compared to 11.9 and 12.7% in pregnant women without cholestasis and healthy Caucasian controls, respectively (Table 4). Furthermore, one CIC patient was a heterozygous carrier for the two variant alleles at positions 3563 and 4544.

For correlation of bile acid levels with the corresponding genotype at position 1331 of ABCB11, ICP and CIC samples were analyzed together. Bile acid levels were available for 16 out of 21 ICP_new patients, seven out of 20 ICP_old patients, and three out of four CIC patients, which yielded a total of 26 samples (CC, 14 patients; CT, 10 patients; and TT, two patients). Interindividual variability in serum bile acid levels was high and ranged from 1.7 to 22.3 ULN and 1.7 to 17.3 ULN in CC and CT patients, respectively. Serum bile acid levels gradually increased from carriers of the TT genotype to carriers of the CC genotype, with medians of 2.3 ULN (Q1, 2.2; Q3, 2.5), 3.2 ULN (Q1, 2.4; Q3, 6.2) and 4.4 ULN (Q1, 2.2; Q3, 7.8) for TT, TC and CC, respectively (Figure 2).

Open in New Tab   Full Size Figure   Download Figure

Figure 2 Bile acid levels in patients harboring different 1331C>T genotypes. TT: two patients; CT: 10 patients; CC: 14 patients.

Intrahepatic cholestasis of pregnancy (ICP) and oral contraceptive-induced cholestasis (CIC) are two acquired forms of cholestasis, which are observed in otherwise healthy young women with a normal medical history. The bile salt export pump (BSEP, ABCB11) and the multidrug resistance protein 2 (MRP2, ABCC2) might be of pathogenetic importance in both conditions.

A genetic predisposition for both types of hormonal cholestasis has been suspected based upon the strong regional clustering, the higher prevalence in female family members of patients with ICP, and the co-incidence with hereditary cases of progressive familial intrahepatic cholestasis. While mutations in the ABCB4 gene that encodes the canalicular phospholipid flippase multidrug resistance protein 3 (MDR3) have been implicated in the development of ICP and CIC in a subset of affected patients, the role of genetic variants in ABCB11 and ABCC2 remains unclear.

Our data support a role of the ABCB11 1331T>C polymorphism as a susceptibility factor for the development of estrogen-induced cholestasis, whereas no such association was found for ABCC2. Serum bile acid and γ-GT levels might help to distinguish ABCB4- and ABCB11-related forms of ICP and CIC.

While the clinical consequences of such findings are still uncertain at this time, they provide important new insights in the role of genetically determined differences in canalicular transporter expression and function for the development of estrogen-induced cholestasis. In the future, the integration of different factors that predict cholestasis might be used to counsel pregnant patients or to avoid certain medications in susceptible patients.

ICP: Intrahepatic cholestasis of pregnancy; CIC: contraceptive-induced cholestasis; BSEP: Bile Salt Export Pump (ABCB11); MRP2: Multidrug Resistance Protein 2 (ABCC2); MDR3: Multidrug Resistance Protein 3 (ABCB4).

The study characterized a potential underlying defect in the subgroup of normal γ-GT ICP patients and contributes to a clinical risk assessment for the future. This study from a group with longstanding experience in transporter genomics is well designed and presented in a clearly written manuscript.